Studies may include both sexes yet leave differences unexplored. That gap becomes more consequential when old evidence helps train medical AI.